Application of Deep Learning in Aspect-Based Sentiment Analysis from E-Commerce User Reviews

In the context of the rapid growth of e-commerce, understanding customer emotions and needs through online feedback has become a key factor in strategies to improve product and service quality. On platforms such as Shopee, users often leave reviews after experiencing products - these comments not only reflect customer satisfaction but also provide detailed information regarding various aspects such as price, quality, shipping, and customer service.This study leverages over 11,000 customer reviews on many shoe-related products from the e-commerce platform - Shopee, annotated across 8 different aspects and classified into 4 sentiment categories: positive, negative, neutral, and undefined. The main objective is to perform Aspect-Based Sentiment Analysis (ABSA) to uncover insights into user experiences and propose improvements to products and services.By combining advanced text preprocessing techniques, data visualization, and modern machine learning models, this research not only evaluates overall sentiment but also analyzes detailed factors influencing consumer purchasing decisions and satisfaction. This study contributes to the development of Vietnamese natural language processing and opens up future research directions for low-resource languages.